Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Diffusion training #13

Merged
merged 6 commits into from
Apr 4, 2024
Merged

Diffusion training #13

merged 6 commits into from
Apr 4, 2024

Conversation

sblackburn86
Copy link
Collaborator

Changing the train.py entry point to handle a diffusion model instead in train_diffusion.py.
This trains the model end-to-end on real data!
examples/local/run_diffusion.sh gives an example on how to do this

Reorganization of the config file to handle the different arguments as cleanly as possible. We should check the library simple_parsing in the future to clean up even more.

Rewrite the load_model function - some cleaning will be required in the future if we integrate simple_parsing.

Harmonize the names of variables between the dataloader and the model.

Training a model requires about 45 minutes of preprocessing at the moment. This is only done once - to create .parquet file that are parsed by the library Datasets.

TODO: integrate orion and more metrics in tensorboard

spatial_dimension: 3
model:
score_network:
hidden_dimensions: [16, 16] # dimensions of the hidden layers. Length of array determines number of la
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The comment seems to have been truncated. number of layers?

Copy link
Collaborator

@rousseab rousseab left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM!

@sblackburn86 sblackburn86 merged commit c8eab3d into main Apr 4, 2024
1 check passed
@sblackburn86 sblackburn86 deleted the diffusion_training branch April 4, 2024 14:34
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants